Using sampling techniques to obtain demographic and social information on the immigrant population requires a specifically designed survey. To guarantee the reliability of results, the sample design should not be limited to commonly used demographic variables, but should also consider the immigrants’ year of arrival. If this variable is not included in the initial design or subsequent adjustments of the survey, there is a high risk that its estimate will be biased, thereby affecting the estimates of all variables correlated with the year of arrival.
The paper analyses the Spanish case for two reasons. Firstly, because Spain has an administrative source – the Municipal Population Register – which records migrants irrespective of their legal situation, so the figures it provides may be a good reference for comparing the estimates generated from the surveys. Secondly, in addition to the migration information provided by the Spanish Labour Force Survey, data from a new specific survey of the immigration population – the National Immigrants Survey-2007 – were also available to us. The existence of these three separate sources makes Spain the best testing ground in Europe for evaluating the reliability of sampling techniques to measure migration flows. Although the analysis is carried out essentially with Spanish statistical information, it will be shown that the problems are also common to other European countries.